From: Pavan Nikhilesh Bhagavatula <pbhagavatula@marvell.com>
To: Apeksha Gupta <apeksha.gupta@nxp.com>,
"jerin.jacob@caviumnetworks.com" <jerin.jacob@caviumnetworks.com>
Cc: "dev@dpdk.org" <dev@dpdk.org>,
"thomas@monjalon.net" <thomas@monjalon.net>,
"hemant.agrawal@nxp.com" <hemant.agrawal@nxp.com>,
"nipun.gupta@nxp.com" <nipun.gupta@nxp.com>,
"akhil.goyal@nxp.com" <akhil.goyal@nxp.com>,
"stable@dpdk.org" <stable@dpdk.org>
Subject: Re: [dpdk-dev] [PATCH 1/6] app/test-eventdev: Enhancing perf-queue packet flow
Date: Thu, 2 Jul 2020 03:24:07 +0000 [thread overview]
Message-ID: <BN6PR18MB1140EA41EFD88D4D05233C93DE6D0@BN6PR18MB1140.namprd18.prod.outlook.com> (raw)
In-Reply-To: <20200701060626.28627-1-apeksha.gupta@nxp.com>
>Subject: [dpdk-dev] [PATCH 1/6] app/test-eventdev: Enhancing perf-
>queue packet flow
>
>The event ethernet Tx adapter provides data path for the ethernet
>transmit
>stage. Enqueue a burst of events objects supplied on an event device.
>
NAK, please use pipeline_atq/queue to test Rx->Tx performance.
Perf_atq/queue should only be used to test event device performance/latency such as
<event_src(CPU/Rx/timer)> -> worker.
>Fixes: 2369f73329 ("app/testeventdev: add perf queue worker
>functions")
>Cc: stable@dpdk.org
>
>Signed-off-by: Apeksha Gupta <apeksha.gupta@nxp.com>
>---
> app/test-eventdev/test_perf_common.c | 11 ++++++++
> app/test-eventdev/test_perf_common.h | 1 +
> app/test-eventdev/test_perf_queue.c | 42 ++++++++++++++++++++---
>-----
> 3 files changed, 43 insertions(+), 11 deletions(-)
>
>diff --git a/app/test-eventdev/test_perf_common.c b/app/test-
>eventdev/test_perf_common.c
>index b3af4bfeca..341e16eade 100644
>--- a/app/test-eventdev/test_perf_common.c
>+++ b/app/test-eventdev/test_perf_common.c
>@@ -687,9 +687,20 @@ perf_ethdev_setup(struct evt_test *test,
>struct evt_options *opt)
> return -ENODEV;
> }
>
>+ t->internal_port = 1;
> RTE_ETH_FOREACH_DEV(i) {
> struct rte_eth_dev_info dev_info;
> struct rte_eth_conf local_port_conf = port_conf;
>+ uint32_t caps = 0;
>+
>+ ret = rte_event_eth_tx_adapter_caps_get(opt->dev_id,
>i, &caps);
>+ if (ret != 0) {
>+ evt_err("failed to get event tx adapter[%d]
>caps", i);
>+ return ret;
>+ }
>+
>+ if (!(caps &
>RTE_EVENT_ETH_TX_ADAPTER_CAP_INTERNAL_PORT))
>+ t->internal_port = 0;
>
> ret = rte_eth_dev_info_get(i, &dev_info);
> if (ret != 0) {
>diff --git a/app/test-eventdev/test_perf_common.h b/app/test-
>eventdev/test_perf_common.h
>index d8fbee6d89..716199d8c9 100644
>--- a/app/test-eventdev/test_perf_common.h
>+++ b/app/test-eventdev/test_perf_common.h
>@@ -48,6 +48,7 @@ struct test_perf {
> int done;
> uint64_t outstand_pkts;
> uint8_t nb_workers;
>+ uint8_t internal_port;
> enum evt_test_result result;
> uint32_t nb_flows;
> uint64_t nb_pkts;
>diff --git a/app/test-eventdev/test_perf_queue.c b/app/test-
>eventdev/test_perf_queue.c
>index 29098580e7..f79e4a4164 100644
>--- a/app/test-eventdev/test_perf_queue.c
>+++ b/app/test-eventdev/test_perf_queue.c
>@@ -71,10 +71,12 @@ perf_queue_worker(void *arg, const int
>enable_fwd_latency)
> }
>
> static int
>-perf_queue_worker_burst(void *arg, const int enable_fwd_latency)
>+perf_queue_worker_burst(void *arg, const int enable_fwd_latency,
>+ const uint32_t flags)
> {
> PERF_WORKER_INIT;
> uint16_t i;
>+ uint16_t nb_tx;
> /* +1 to avoid prefetch out of array check */
> struct rte_event ev[BURST_SIZE + 1];
>
>@@ -111,12 +113,20 @@ perf_queue_worker_burst(void *arg, const
>int enable_fwd_latency)
> }
> }
>
>- uint16_t enq;
>-
>- enq = rte_event_enqueue_burst(dev, port, ev, nb_rx);
>- while (enq < nb_rx) {
>- enq += rte_event_enqueue_burst(dev, port,
>+ if (flags == TEST_PERF_EVENT_TX_DIRECT) {
>+ nb_tx =
>rte_event_eth_tx_adapter_enqueue(dev, port,
>+ ev,
>nb_rx, 0);
>+ while (nb_tx < nb_rx && !t->done)
>+ nb_tx +=
>rte_event_eth_tx_adapter_enqueue(dev,
>+ port, ev +
>nb_tx,
>+ nb_rx - nb_tx,
>0);
>+ } else {
>+ uint16_t enq;
>+ enq = rte_event_enqueue_burst(dev, port, ev,
>nb_rx);
>+ while (enq < nb_rx) {
>+ enq += rte_event_enqueue_burst(dev,
>port,
> ev + enq, nb_rx
>- enq);
>+ }
> }
> }
> return 0;
>@@ -130,16 +140,26 @@ worker_wrapper(void *arg)
>
> const bool burst = evt_has_burst_mode(w->dev_id);
> const int fwd_latency = opt->fwd_latency;
>-
>+ const bool internal_port = w->t->internal_port;
>+ uint32_t flags;
> /* allow compiler to optimize */
> if (!burst && !fwd_latency)
> return perf_queue_worker(arg, 0);
> else if (!burst && fwd_latency)
> return perf_queue_worker(arg, 1);
>- else if (burst && !fwd_latency)
>- return perf_queue_worker_burst(arg, 0);
>- else if (burst && fwd_latency)
>- return perf_queue_worker_burst(arg, 1);
>+ else if (burst && !fwd_latency && internal_port) {
>+ flags = TEST_PERF_EVENT_TX_DIRECT;
>+ return perf_queue_worker_burst(arg, 0, flags);
>+ } else if (burst && !fwd_latency && !internal_port) {
>+ flags = TEST_PERF_EVENT_TX_ENQ;
>+ return perf_queue_worker_burst(arg, 1, flags);
>+ } else if (burst && fwd_latency && internal_port) {
>+ flags = TEST_PERF_EVENT_TX_DIRECT;
>+ return perf_queue_worker_burst(arg, 0, flags);
>+ } else if (burst && fwd_latency && !internal_port) {
>+ flags = TEST_PERF_EVENT_TX_ENQ;
>+ return perf_queue_worker_burst(arg, 1, flags);
>+ }
>
> rte_panic("invalid worker\n");
> }
>--
>2.17.1
next prev parent reply other threads:[~2020-07-02 3:24 UTC|newest]
Thread overview: 22+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-07-01 6:06 Apeksha Gupta
2020-07-01 6:06 ` [dpdk-dev] [PATCH 2/6] app/test-eventdev: Removed unwanted checks Apeksha Gupta
2020-07-01 6:06 ` [dpdk-dev] [PATCH 3/6] event/dpaa2: Add all type queue capability flag Apeksha Gupta
2020-07-01 6:06 ` [dpdk-dev] [PATCH 4/6] app/test-eventdev: Fix pipeline atq Apeksha Gupta
2020-07-10 13:08 ` [dpdk-dev] [PATCH v2] " Apeksha Gupta
2020-07-13 11:44 ` [dpdk-dev] [PATCH v3 1/2] " Apeksha Gupta
2020-07-13 11:44 ` [dpdk-dev] [PATCH v3 2/2] event/dpaa2: Add all type queue capability flag Apeksha Gupta
2020-07-21 17:51 ` Jerin Jacob
2020-07-23 13:36 ` Nipun Gupta
2020-07-24 5:24 ` Jerin Jacob
2020-07-23 17:04 ` [dpdk-dev] [PATCH v4 2/2] event/dpaa2: add " Apeksha Gupta
2020-07-22 1:13 ` [dpdk-dev] [PATCH v3 1/2] app/test-eventdev: Fix pipeline atq Pavan Nikhilesh Bhagavatula
2020-07-22 7:48 ` Thomas Monjalon
2020-07-23 17:02 ` [dpdk-dev] [PATCH v4 1/2] app/test-eventdev: fix capability check in pipeline ATQ test Apeksha Gupta
2020-07-01 6:06 ` [dpdk-dev] [PATCH 5/6] app/test-eventdev: Enhancing perf-atq packet flow Apeksha Gupta
2020-07-02 3:25 ` Pavan Nikhilesh Bhagavatula
2020-07-02 14:46 ` [dpdk-dev] [EXT] " Apeksha Gupta
2020-07-01 6:06 ` [dpdk-dev] [PATCH 6/6] app/test-eventdev: fix eventdev queues Apeksha Gupta
2020-07-02 3:27 ` Pavan Nikhilesh Bhagavatula
2020-07-09 13:05 ` [dpdk-dev] [EXT] " Apeksha Gupta
2020-07-02 3:24 ` Pavan Nikhilesh Bhagavatula [this message]
2020-07-02 14:46 ` [dpdk-dev] [EXT] RE: [PATCH 1/6] app/test-eventdev: Enhancing perf-queue packet flow Apeksha Gupta
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=BN6PR18MB1140EA41EFD88D4D05233C93DE6D0@BN6PR18MB1140.namprd18.prod.outlook.com \
--to=pbhagavatula@marvell.com \
--cc=akhil.goyal@nxp.com \
--cc=apeksha.gupta@nxp.com \
--cc=dev@dpdk.org \
--cc=hemant.agrawal@nxp.com \
--cc=jerin.jacob@caviumnetworks.com \
--cc=nipun.gupta@nxp.com \
--cc=stable@dpdk.org \
--cc=thomas@monjalon.net \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).